4 * Copyright (c) 2012 Samsung Electronics Co., Ltd.
5 * http://www.samsung.com/
7 * This program is free software; you can redistribute it and/or modify
8 * it under the terms of the GNU General Public License version 2 as
9 * published by the Free Software Foundation.
14 #include <linux/types.h>
15 #include <linux/page-flags.h>
16 #include <linux/buffer_head.h>
17 #include <linux/slab.h>
18 #include <linux/crc32.h>
19 #include <linux/magic.h>
20 #include <linux/kobject.h>
21 #include <linux/sched.h>
23 #ifdef CONFIG_F2FS_CHECK_FS
24 #define f2fs_bug_on(condition) BUG_ON(condition)
26 #define f2fs_bug_on(condition)
32 #define F2FS_MOUNT_BG_GC 0x00000001
33 #define F2FS_MOUNT_DISABLE_ROLL_FORWARD 0x00000002
34 #define F2FS_MOUNT_DISCARD 0x00000004
35 #define F2FS_MOUNT_NOHEAP 0x00000008
36 #define F2FS_MOUNT_XATTR_USER 0x00000010
37 #define F2FS_MOUNT_POSIX_ACL 0x00000020
38 #define F2FS_MOUNT_DISABLE_EXT_IDENTIFY 0x00000040
39 #define F2FS_MOUNT_INLINE_XATTR 0x00000080
41 #define clear_opt(sbi, option) (sbi->mount_opt.opt &= ~F2FS_MOUNT_##option)
42 #define set_opt(sbi, option) (sbi->mount_opt.opt |= F2FS_MOUNT_##option)
43 #define test_opt(sbi, option) (sbi->mount_opt.opt & F2FS_MOUNT_##option)
45 #define ver_after(a, b) (typecheck(unsigned long long, a) && \
46 typecheck(unsigned long long, b) && \
47 ((long long)((a) - (b)) > 0))
49 typedef u32 block_t
; /*
50 * should not change u32, since it is the on-disk block
51 * address format, __le32.
55 struct f2fs_mount_info
{
59 #define CRCPOLY_LE 0xedb88320
61 static inline __u32
f2fs_crc32(void *buf
, size_t len
)
63 unsigned char *p
= (unsigned char *)buf
;
64 __u32 crc
= F2FS_SUPER_MAGIC
;
69 for (i
= 0; i
< 8; i
++)
70 crc
= (crc
>> 1) ^ ((crc
& 1) ? CRCPOLY_LE
: 0);
75 static inline bool f2fs_crc_valid(__u32 blk_crc
, void *buf
, size_t buf_size
)
77 return f2fs_crc32(buf
, buf_size
) == blk_crc
;
81 * For checkpoint manager
88 /* for the list of orphan inodes */
89 struct orphan_inode_entry
{
90 struct list_head list
; /* list head */
91 nid_t ino
; /* inode number */
94 /* for the list of directory inodes */
95 struct dir_inode_entry
{
96 struct list_head list
; /* list head */
97 struct inode
*inode
; /* vfs inode pointer */
100 /* for the list of blockaddresses to be discarded */
101 struct discard_entry
{
102 struct list_head list
; /* list head */
103 block_t blkaddr
; /* block address to be discarded */
104 int len
; /* # of consecutive blocks of the discard */
107 /* for the list of fsync inodes, used only during recovery */
108 struct fsync_inode_entry
{
109 struct list_head list
; /* list head */
110 struct inode
*inode
; /* vfs inode pointer */
111 block_t blkaddr
; /* block address locating the last inode */
114 #define nats_in_cursum(sum) (le16_to_cpu(sum->n_nats))
115 #define sits_in_cursum(sum) (le16_to_cpu(sum->n_sits))
117 #define nat_in_journal(sum, i) (sum->nat_j.entries[i].ne)
118 #define nid_in_journal(sum, i) (sum->nat_j.entries[i].nid)
119 #define sit_in_journal(sum, i) (sum->sit_j.entries[i].se)
120 #define segno_in_journal(sum, i) (sum->sit_j.entries[i].segno)
122 static inline int update_nats_in_cursum(struct f2fs_summary_block
*rs
, int i
)
124 int before
= nats_in_cursum(rs
);
125 rs
->n_nats
= cpu_to_le16(before
+ i
);
129 static inline int update_sits_in_cursum(struct f2fs_summary_block
*rs
, int i
)
131 int before
= sits_in_cursum(rs
);
132 rs
->n_sits
= cpu_to_le16(before
+ i
);
139 #define F2FS_IOC_GETFLAGS FS_IOC_GETFLAGS
140 #define F2FS_IOC_SETFLAGS FS_IOC_SETFLAGS
142 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
144 * ioctl commands in 32 bit emulation
146 #define F2FS_IOC32_GETFLAGS FS_IOC32_GETFLAGS
147 #define F2FS_IOC32_SETFLAGS FS_IOC32_SETFLAGS
151 * For INODE and NODE manager
154 * XATTR_NODE_OFFSET stores xattrs to one node block per file keeping -1
155 * as its node offset to distinguish from index node blocks.
156 * But some bits are used to mark the node block.
158 #define XATTR_NODE_OFFSET ((((unsigned int)-1) << OFFSET_BIT_SHIFT) \
161 ALLOC_NODE
, /* allocate a new node page if needed */
162 LOOKUP_NODE
, /* look up a node without readahead */
164 * look up a node with readahead called
165 * by get_datablock_ro.
169 #define F2FS_LINK_MAX 32000 /* maximum link count per file */
171 /* for in-memory extent cache entry */
173 rwlock_t ext_lock
; /* rwlock for consistency */
174 unsigned int fofs
; /* start offset in a file */
175 u32 blk_addr
; /* start block address of the extent */
176 unsigned int len
; /* length of the extent */
180 * i_advise uses FADVISE_XXX_BIT. We can add additional hints later.
182 #define FADVISE_COLD_BIT 0x01
183 #define FADVISE_LOST_PINO_BIT 0x02
185 struct f2fs_inode_info
{
186 struct inode vfs_inode
; /* serve a vfs inode */
187 unsigned long i_flags
; /* keep an inode flags for ioctl */
188 unsigned char i_advise
; /* use to give file attribute hints */
189 unsigned int i_current_depth
; /* use only in directory structure */
190 unsigned int i_pino
; /* parent inode number */
191 umode_t i_acl_mode
; /* keep file acl mode temporarily */
193 /* Use below internally in f2fs*/
194 unsigned long flags
; /* use to pass per-file flags */
195 atomic_t dirty_dents
; /* # of dirty dentry pages */
196 f2fs_hash_t chash
; /* hash value of given file name */
197 unsigned int clevel
; /* maximum level of given file name */
198 nid_t i_xattr_nid
; /* node id that contains xattrs */
199 unsigned long long xattr_ver
; /* cp version of xattr modification */
200 struct extent_info ext
; /* in-memory extent cache entry */
203 static inline void get_extent_info(struct extent_info
*ext
,
204 struct f2fs_extent i_ext
)
206 write_lock(&ext
->ext_lock
);
207 ext
->fofs
= le32_to_cpu(i_ext
.fofs
);
208 ext
->blk_addr
= le32_to_cpu(i_ext
.blk_addr
);
209 ext
->len
= le32_to_cpu(i_ext
.len
);
210 write_unlock(&ext
->ext_lock
);
213 static inline void set_raw_extent(struct extent_info
*ext
,
214 struct f2fs_extent
*i_ext
)
216 read_lock(&ext
->ext_lock
);
217 i_ext
->fofs
= cpu_to_le32(ext
->fofs
);
218 i_ext
->blk_addr
= cpu_to_le32(ext
->blk_addr
);
219 i_ext
->len
= cpu_to_le32(ext
->len
);
220 read_unlock(&ext
->ext_lock
);
223 struct f2fs_nm_info
{
224 block_t nat_blkaddr
; /* base disk address of NAT */
225 nid_t max_nid
; /* maximum possible node ids */
226 nid_t next_scan_nid
; /* the next nid to be scanned */
228 /* NAT cache management */
229 struct radix_tree_root nat_root
;/* root of the nat entry cache */
230 rwlock_t nat_tree_lock
; /* protect nat_tree_lock */
231 unsigned int nat_cnt
; /* the # of cached nat entries */
232 struct list_head nat_entries
; /* cached nat entry list (clean) */
233 struct list_head dirty_nat_entries
; /* cached nat entry list (dirty) */
235 /* free node ids management */
236 struct list_head free_nid_list
; /* a list for free nids */
237 spinlock_t free_nid_list_lock
; /* protect free nid list */
238 unsigned int fcnt
; /* the number of free node id */
239 struct mutex build_lock
; /* lock for build free nids */
242 char *nat_bitmap
; /* NAT bitmap pointer */
243 int bitmap_size
; /* bitmap size */
247 * this structure is used as one of function parameters.
248 * all the information are dedicated to a given direct node block determined
249 * by the data offset in a file.
251 struct dnode_of_data
{
252 struct inode
*inode
; /* vfs inode pointer */
253 struct page
*inode_page
; /* its inode page, NULL is possible */
254 struct page
*node_page
; /* cached direct node page */
255 nid_t nid
; /* node id of the direct node block */
256 unsigned int ofs_in_node
; /* data offset in the node page */
257 bool inode_page_locked
; /* inode page is locked or not */
258 block_t data_blkaddr
; /* block address of the node block */
261 static inline void set_new_dnode(struct dnode_of_data
*dn
, struct inode
*inode
,
262 struct page
*ipage
, struct page
*npage
, nid_t nid
)
264 memset(dn
, 0, sizeof(*dn
));
266 dn
->inode_page
= ipage
;
267 dn
->node_page
= npage
;
274 * By default, there are 6 active log areas across the whole main area.
275 * When considering hot and cold data separation to reduce cleaning overhead,
276 * we split 3 for data logs and 3 for node logs as hot, warm, and cold types,
278 * In the current design, you should not change the numbers intentionally.
279 * Instead, as a mount option such as active_logs=x, you can use 2, 4, and 6
280 * logs individually according to the underlying devices. (default: 6)
281 * Just in case, on-disk layout covers maximum 16 logs that consist of 8 for
282 * data and 8 for node logs.
284 #define NR_CURSEG_DATA_TYPE (3)
285 #define NR_CURSEG_NODE_TYPE (3)
286 #define NR_CURSEG_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE)
289 CURSEG_HOT_DATA
= 0, /* directory entry blocks */
290 CURSEG_WARM_DATA
, /* data blocks */
291 CURSEG_COLD_DATA
, /* multimedia or GCed data blocks */
292 CURSEG_HOT_NODE
, /* direct node blocks of directory files */
293 CURSEG_WARM_NODE
, /* direct node blocks of normal files */
294 CURSEG_COLD_NODE
, /* indirect node blocks */
298 struct f2fs_sm_info
{
299 struct sit_info
*sit_info
; /* whole segment information */
300 struct free_segmap_info
*free_info
; /* free segment information */
301 struct dirty_seglist_info
*dirty_info
; /* dirty segment information */
302 struct curseg_info
*curseg_array
; /* active segment information */
304 struct list_head wblist_head
; /* list of under-writeback pages */
305 spinlock_t wblist_lock
; /* lock for checkpoint */
307 block_t seg0_blkaddr
; /* block address of 0'th segment */
308 block_t main_blkaddr
; /* start block address of main area */
309 block_t ssa_blkaddr
; /* start block address of SSA area */
311 unsigned int segment_count
; /* total # of segments */
312 unsigned int main_segments
; /* # of segments in main area */
313 unsigned int reserved_segments
; /* # of reserved segments */
314 unsigned int ovp_segments
; /* # of overprovision segments */
316 /* a threshold to reclaim prefree segments */
317 unsigned int rec_prefree_segments
;
319 /* for small discard management */
320 struct list_head discard_list
; /* 4KB discard list */
321 int nr_discards
; /* # of discards in the list */
322 int max_discards
; /* max. discards to be issued */
329 * COUNT_TYPE for monitoring
331 * f2fs monitors the number of several block types such as on-writeback,
332 * dirty dentry blocks, dirty node blocks, and dirty meta blocks.
343 * The below are the page types of bios used in submti_bio().
344 * The available types are:
345 * DATA User data pages. It operates as async mode.
346 * NODE Node pages. It operates as async mode.
347 * META FS metadata pages such as SIT, NAT, CP.
348 * NR_PAGE_TYPE The number of page types.
349 * META_FLUSH Make sure the previous pages are written
350 * with waiting the bio's completion
351 * ... Only can be used with META.
353 #define PAGE_TYPE_OF_BIO(type) ((type) > META ? META : (type))
362 struct f2fs_sb_info
{
363 struct super_block
*sb
; /* pointer to VFS super block */
364 struct proc_dir_entry
*s_proc
; /* proc entry */
365 struct buffer_head
*raw_super_buf
; /* buffer head of raw sb */
366 struct f2fs_super_block
*raw_super
; /* raw super block pointer */
367 int s_dirty
; /* dirty flag for checkpoint */
369 /* for node-related operations */
370 struct f2fs_nm_info
*nm_info
; /* node manager */
371 struct inode
*node_inode
; /* cache node blocks */
373 /* for segment-related operations */
374 struct f2fs_sm_info
*sm_info
; /* segment manager */
375 struct bio
*bio
[NR_PAGE_TYPE
]; /* bios to merge */
376 sector_t last_block_in_bio
[NR_PAGE_TYPE
]; /* last block number */
377 struct mutex write_mutex
[NR_PAGE_TYPE
]; /* mutex for writing IOs */
380 struct f2fs_checkpoint
*ckpt
; /* raw checkpoint pointer */
381 struct inode
*meta_inode
; /* cache meta blocks */
382 struct mutex cp_mutex
; /* checkpoint procedure lock */
383 struct rw_semaphore cp_rwsem
; /* blocking FS operations */
384 struct mutex node_write
; /* locking node writes */
385 struct mutex writepages
; /* mutex for writepages() */
386 bool por_doing
; /* recovery is doing or not */
387 bool on_build_free_nids
; /* build_free_nids is doing */
388 wait_queue_head_t cp_wait
;
390 /* for orphan inode management */
391 struct list_head orphan_inode_list
; /* orphan inode list */
392 struct mutex orphan_inode_mutex
; /* for orphan inode list */
393 unsigned int n_orphans
; /* # of orphan inodes */
395 /* for directory inode management */
396 struct list_head dir_inode_list
; /* dir inode list */
397 spinlock_t dir_inode_lock
; /* for dir inode list lock */
399 /* basic file system units */
400 unsigned int log_sectors_per_block
; /* log2 sectors per block */
401 unsigned int log_blocksize
; /* log2 block size */
402 unsigned int blocksize
; /* block size */
403 unsigned int root_ino_num
; /* root inode number*/
404 unsigned int node_ino_num
; /* node inode number*/
405 unsigned int meta_ino_num
; /* meta inode number*/
406 unsigned int log_blocks_per_seg
; /* log2 blocks per segment */
407 unsigned int blocks_per_seg
; /* blocks per segment */
408 unsigned int segs_per_sec
; /* segments per section */
409 unsigned int secs_per_zone
; /* sections per zone */
410 unsigned int total_sections
; /* total section count */
411 unsigned int total_node_count
; /* total node block count */
412 unsigned int total_valid_node_count
; /* valid node block count */
413 unsigned int total_valid_inode_count
; /* valid inode count */
414 int active_logs
; /* # of active logs */
416 block_t user_block_count
; /* # of user blocks */
417 block_t total_valid_block_count
; /* # of valid blocks */
418 block_t alloc_valid_block_count
; /* # of allocated blocks */
419 block_t last_valid_block_count
; /* for recovery */
420 u32 s_next_generation
; /* for NFS support */
421 atomic_t nr_pages
[NR_COUNT_TYPE
]; /* # of pages, see count_type */
423 struct f2fs_mount_info mount_opt
; /* mount options */
425 /* for cleaning operations */
426 struct mutex gc_mutex
; /* mutex for GC */
427 struct f2fs_gc_kthread
*gc_thread
; /* GC thread */
428 unsigned int cur_victim_sec
; /* current victim section num */
431 * for stat information.
432 * one is for the LFS mode, and the other is for the SSR mode.
434 #ifdef CONFIG_F2FS_STAT_FS
435 struct f2fs_stat_info
*stat_info
; /* FS status information */
436 unsigned int segment_count
[2]; /* # of allocated segments */
437 unsigned int block_count
[2]; /* # of allocated blocks */
438 int total_hit_ext
, read_hit_ext
; /* extent cache hit ratio */
439 int bg_gc
; /* background gc calls */
440 unsigned int n_dirty_dirs
; /* # of dir inodes */
442 unsigned int last_victim
[2]; /* last victim segment # */
443 spinlock_t stat_lock
; /* lock for stat operations */
445 /* For sysfs suppport */
446 struct kobject s_kobj
;
447 struct completion s_kobj_unregister
;
453 static inline struct f2fs_inode_info
*F2FS_I(struct inode
*inode
)
455 return container_of(inode
, struct f2fs_inode_info
, vfs_inode
);
458 static inline struct f2fs_sb_info
*F2FS_SB(struct super_block
*sb
)
460 return sb
->s_fs_info
;
463 static inline struct f2fs_super_block
*F2FS_RAW_SUPER(struct f2fs_sb_info
*sbi
)
465 return (struct f2fs_super_block
*)(sbi
->raw_super
);
468 static inline struct f2fs_checkpoint
*F2FS_CKPT(struct f2fs_sb_info
*sbi
)
470 return (struct f2fs_checkpoint
*)(sbi
->ckpt
);
473 static inline struct f2fs_node
*F2FS_NODE(struct page
*page
)
475 return (struct f2fs_node
*)page_address(page
);
478 static inline struct f2fs_nm_info
*NM_I(struct f2fs_sb_info
*sbi
)
480 return (struct f2fs_nm_info
*)(sbi
->nm_info
);
483 static inline struct f2fs_sm_info
*SM_I(struct f2fs_sb_info
*sbi
)
485 return (struct f2fs_sm_info
*)(sbi
->sm_info
);
488 static inline struct sit_info
*SIT_I(struct f2fs_sb_info
*sbi
)
490 return (struct sit_info
*)(SM_I(sbi
)->sit_info
);
493 static inline struct free_segmap_info
*FREE_I(struct f2fs_sb_info
*sbi
)
495 return (struct free_segmap_info
*)(SM_I(sbi
)->free_info
);
498 static inline struct dirty_seglist_info
*DIRTY_I(struct f2fs_sb_info
*sbi
)
500 return (struct dirty_seglist_info
*)(SM_I(sbi
)->dirty_info
);
503 static inline void F2FS_SET_SB_DIRT(struct f2fs_sb_info
*sbi
)
508 static inline void F2FS_RESET_SB_DIRT(struct f2fs_sb_info
*sbi
)
513 static inline unsigned long long cur_cp_version(struct f2fs_checkpoint
*cp
)
515 return le64_to_cpu(cp
->checkpoint_ver
);
518 static inline bool is_set_ckpt_flags(struct f2fs_checkpoint
*cp
, unsigned int f
)
520 unsigned int ckpt_flags
= le32_to_cpu(cp
->ckpt_flags
);
521 return ckpt_flags
& f
;
524 static inline void set_ckpt_flags(struct f2fs_checkpoint
*cp
, unsigned int f
)
526 unsigned int ckpt_flags
= le32_to_cpu(cp
->ckpt_flags
);
528 cp
->ckpt_flags
= cpu_to_le32(ckpt_flags
);
531 static inline void clear_ckpt_flags(struct f2fs_checkpoint
*cp
, unsigned int f
)
533 unsigned int ckpt_flags
= le32_to_cpu(cp
->ckpt_flags
);
535 cp
->ckpt_flags
= cpu_to_le32(ckpt_flags
);
538 static inline void f2fs_lock_op(struct f2fs_sb_info
*sbi
)
540 down_read(&sbi
->cp_rwsem
);
543 static inline void f2fs_unlock_op(struct f2fs_sb_info
*sbi
)
545 up_read(&sbi
->cp_rwsem
);
548 static inline void f2fs_lock_all(struct f2fs_sb_info
*sbi
)
550 down_write_nest_lock(&sbi
->cp_rwsem
, &sbi
->cp_mutex
);
553 static inline void f2fs_unlock_all(struct f2fs_sb_info
*sbi
)
555 up_write(&sbi
->cp_rwsem
);
559 * Check whether the given nid is within node id range.
561 static inline int check_nid_range(struct f2fs_sb_info
*sbi
, nid_t nid
)
563 WARN_ON((nid
>= NM_I(sbi
)->max_nid
));
564 if (nid
>= NM_I(sbi
)->max_nid
)
569 #define F2FS_DEFAULT_ALLOCATED_BLOCKS 1
572 * Check whether the inode has blocks or not
574 static inline int F2FS_HAS_BLOCKS(struct inode
*inode
)
576 if (F2FS_I(inode
)->i_xattr_nid
)
577 return (inode
->i_blocks
> F2FS_DEFAULT_ALLOCATED_BLOCKS
+ 1);
579 return (inode
->i_blocks
> F2FS_DEFAULT_ALLOCATED_BLOCKS
);
582 static inline bool inc_valid_block_count(struct f2fs_sb_info
*sbi
,
583 struct inode
*inode
, blkcnt_t count
)
585 block_t valid_block_count
;
587 spin_lock(&sbi
->stat_lock
);
589 sbi
->total_valid_block_count
+ (block_t
)count
;
590 if (valid_block_count
> sbi
->user_block_count
) {
591 spin_unlock(&sbi
->stat_lock
);
594 inode
->i_blocks
+= count
;
595 sbi
->total_valid_block_count
= valid_block_count
;
596 sbi
->alloc_valid_block_count
+= (block_t
)count
;
597 spin_unlock(&sbi
->stat_lock
);
601 static inline int dec_valid_block_count(struct f2fs_sb_info
*sbi
,
605 spin_lock(&sbi
->stat_lock
);
606 f2fs_bug_on(sbi
->total_valid_block_count
< (block_t
) count
);
607 f2fs_bug_on(inode
->i_blocks
< count
);
608 inode
->i_blocks
-= count
;
609 sbi
->total_valid_block_count
-= (block_t
)count
;
610 spin_unlock(&sbi
->stat_lock
);
614 static inline void inc_page_count(struct f2fs_sb_info
*sbi
, int count_type
)
616 atomic_inc(&sbi
->nr_pages
[count_type
]);
617 F2FS_SET_SB_DIRT(sbi
);
620 static inline void inode_inc_dirty_dents(struct inode
*inode
)
622 atomic_inc(&F2FS_I(inode
)->dirty_dents
);
625 static inline void dec_page_count(struct f2fs_sb_info
*sbi
, int count_type
)
627 atomic_dec(&sbi
->nr_pages
[count_type
]);
630 static inline void inode_dec_dirty_dents(struct inode
*inode
)
632 atomic_dec(&F2FS_I(inode
)->dirty_dents
);
635 static inline int get_pages(struct f2fs_sb_info
*sbi
, int count_type
)
637 return atomic_read(&sbi
->nr_pages
[count_type
]);
640 static inline int get_blocktype_secs(struct f2fs_sb_info
*sbi
, int block_type
)
642 unsigned int pages_per_sec
= sbi
->segs_per_sec
*
643 (1 << sbi
->log_blocks_per_seg
);
644 return ((get_pages(sbi
, block_type
) + pages_per_sec
- 1)
645 >> sbi
->log_blocks_per_seg
) / sbi
->segs_per_sec
;
648 static inline block_t
valid_user_blocks(struct f2fs_sb_info
*sbi
)
651 spin_lock(&sbi
->stat_lock
);
652 ret
= sbi
->total_valid_block_count
;
653 spin_unlock(&sbi
->stat_lock
);
657 static inline unsigned long __bitmap_size(struct f2fs_sb_info
*sbi
, int flag
)
659 struct f2fs_checkpoint
*ckpt
= F2FS_CKPT(sbi
);
661 /* return NAT or SIT bitmap */
662 if (flag
== NAT_BITMAP
)
663 return le32_to_cpu(ckpt
->nat_ver_bitmap_bytesize
);
664 else if (flag
== SIT_BITMAP
)
665 return le32_to_cpu(ckpt
->sit_ver_bitmap_bytesize
);
670 static inline void *__bitmap_ptr(struct f2fs_sb_info
*sbi
, int flag
)
672 struct f2fs_checkpoint
*ckpt
= F2FS_CKPT(sbi
);
673 int offset
= (flag
== NAT_BITMAP
) ?
674 le32_to_cpu(ckpt
->sit_ver_bitmap_bytesize
) : 0;
675 return &ckpt
->sit_nat_version_bitmap
+ offset
;
678 static inline block_t
__start_cp_addr(struct f2fs_sb_info
*sbi
)
681 struct f2fs_checkpoint
*ckpt
= F2FS_CKPT(sbi
);
682 unsigned long long ckpt_version
= cur_cp_version(ckpt
);
684 start_addr
= le32_to_cpu(F2FS_RAW_SUPER(sbi
)->cp_blkaddr
);
687 * odd numbered checkpoint should at cp segment 0
688 * and even segent must be at cp segment 1
690 if (!(ckpt_version
& 1))
691 start_addr
+= sbi
->blocks_per_seg
;
696 static inline block_t
__start_sum_addr(struct f2fs_sb_info
*sbi
)
698 return le32_to_cpu(F2FS_CKPT(sbi
)->cp_pack_start_sum
);
701 static inline bool inc_valid_node_count(struct f2fs_sb_info
*sbi
,
705 block_t valid_block_count
;
706 unsigned int valid_node_count
;
708 spin_lock(&sbi
->stat_lock
);
710 valid_block_count
= sbi
->total_valid_block_count
+ (block_t
)count
;
711 sbi
->alloc_valid_block_count
+= (block_t
)count
;
712 valid_node_count
= sbi
->total_valid_node_count
+ count
;
714 if (valid_block_count
> sbi
->user_block_count
) {
715 spin_unlock(&sbi
->stat_lock
);
719 if (valid_node_count
> sbi
->total_node_count
) {
720 spin_unlock(&sbi
->stat_lock
);
725 inode
->i_blocks
+= count
;
726 sbi
->total_valid_node_count
= valid_node_count
;
727 sbi
->total_valid_block_count
= valid_block_count
;
728 spin_unlock(&sbi
->stat_lock
);
733 static inline void dec_valid_node_count(struct f2fs_sb_info
*sbi
,
737 spin_lock(&sbi
->stat_lock
);
739 f2fs_bug_on(sbi
->total_valid_block_count
< count
);
740 f2fs_bug_on(sbi
->total_valid_node_count
< count
);
741 f2fs_bug_on(inode
->i_blocks
< count
);
743 inode
->i_blocks
-= count
;
744 sbi
->total_valid_node_count
-= count
;
745 sbi
->total_valid_block_count
-= (block_t
)count
;
747 spin_unlock(&sbi
->stat_lock
);
750 static inline unsigned int valid_node_count(struct f2fs_sb_info
*sbi
)
753 spin_lock(&sbi
->stat_lock
);
754 ret
= sbi
->total_valid_node_count
;
755 spin_unlock(&sbi
->stat_lock
);
759 static inline void inc_valid_inode_count(struct f2fs_sb_info
*sbi
)
761 spin_lock(&sbi
->stat_lock
);
762 f2fs_bug_on(sbi
->total_valid_inode_count
== sbi
->total_node_count
);
763 sbi
->total_valid_inode_count
++;
764 spin_unlock(&sbi
->stat_lock
);
767 static inline int dec_valid_inode_count(struct f2fs_sb_info
*sbi
)
769 spin_lock(&sbi
->stat_lock
);
770 f2fs_bug_on(!sbi
->total_valid_inode_count
);
771 sbi
->total_valid_inode_count
--;
772 spin_unlock(&sbi
->stat_lock
);
776 static inline unsigned int valid_inode_count(struct f2fs_sb_info
*sbi
)
779 spin_lock(&sbi
->stat_lock
);
780 ret
= sbi
->total_valid_inode_count
;
781 spin_unlock(&sbi
->stat_lock
);
785 static inline void f2fs_put_page(struct page
*page
, int unlock
)
787 if (!page
|| IS_ERR(page
))
791 f2fs_bug_on(!PageLocked(page
));
794 page_cache_release(page
);
797 static inline void f2fs_put_dnode(struct dnode_of_data
*dn
)
800 f2fs_put_page(dn
->node_page
, 1);
801 if (dn
->inode_page
&& dn
->node_page
!= dn
->inode_page
)
802 f2fs_put_page(dn
->inode_page
, 0);
803 dn
->node_page
= NULL
;
804 dn
->inode_page
= NULL
;
807 static inline struct kmem_cache
*f2fs_kmem_cache_create(const char *name
,
808 size_t size
, void (*ctor
)(void *))
810 return kmem_cache_create(name
, size
, 0, SLAB_RECLAIM_ACCOUNT
, ctor
);
813 static inline void *f2fs_kmem_cache_alloc(struct kmem_cache
*cachep
,
818 entry
= kmem_cache_alloc(cachep
, flags
);
827 #define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino)
829 static inline bool IS_INODE(struct page
*page
)
831 struct f2fs_node
*p
= F2FS_NODE(page
);
832 return RAW_IS_INODE(p
);
835 static inline __le32
*blkaddr_in_node(struct f2fs_node
*node
)
837 return RAW_IS_INODE(node
) ? node
->i
.i_addr
: node
->dn
.addr
;
840 static inline block_t
datablock_addr(struct page
*node_page
,
843 struct f2fs_node
*raw_node
;
845 raw_node
= F2FS_NODE(node_page
);
846 addr_array
= blkaddr_in_node(raw_node
);
847 return le32_to_cpu(addr_array
[offset
]);
850 static inline int f2fs_test_bit(unsigned int nr
, char *addr
)
855 mask
= 1 << (7 - (nr
& 0x07));
859 static inline int f2fs_set_bit(unsigned int nr
, char *addr
)
865 mask
= 1 << (7 - (nr
& 0x07));
871 static inline int f2fs_clear_bit(unsigned int nr
, char *addr
)
877 mask
= 1 << (7 - (nr
& 0x07));
883 /* used for f2fs_inode_info->flags */
885 FI_NEW_INODE
, /* indicate newly allocated inode */
886 FI_DIRTY_INODE
, /* indicate inode is dirty or not */
887 FI_INC_LINK
, /* need to increment i_nlink */
888 FI_ACL_MODE
, /* indicate acl mode */
889 FI_NO_ALLOC
, /* should not allocate any blocks */
890 FI_UPDATE_DIR
, /* should update inode block for consistency */
891 FI_DELAY_IPUT
, /* used for the recovery */
892 FI_INLINE_XATTR
, /* used for inline xattr */
895 static inline void set_inode_flag(struct f2fs_inode_info
*fi
, int flag
)
897 set_bit(flag
, &fi
->flags
);
900 static inline int is_inode_flag_set(struct f2fs_inode_info
*fi
, int flag
)
902 return test_bit(flag
, &fi
->flags
);
905 static inline void clear_inode_flag(struct f2fs_inode_info
*fi
, int flag
)
907 clear_bit(flag
, &fi
->flags
);
910 static inline void set_acl_inode(struct f2fs_inode_info
*fi
, umode_t mode
)
912 fi
->i_acl_mode
= mode
;
913 set_inode_flag(fi
, FI_ACL_MODE
);
916 static inline int cond_clear_inode_flag(struct f2fs_inode_info
*fi
, int flag
)
918 if (is_inode_flag_set(fi
, FI_ACL_MODE
)) {
919 clear_inode_flag(fi
, FI_ACL_MODE
);
925 static inline void get_inline_info(struct f2fs_inode_info
*fi
,
926 struct f2fs_inode
*ri
)
928 if (ri
->i_inline
& F2FS_INLINE_XATTR
)
929 set_inode_flag(fi
, FI_INLINE_XATTR
);
932 static inline void set_raw_inline(struct f2fs_inode_info
*fi
,
933 struct f2fs_inode
*ri
)
937 if (is_inode_flag_set(fi
, FI_INLINE_XATTR
))
938 ri
->i_inline
|= F2FS_INLINE_XATTR
;
941 static inline unsigned int addrs_per_inode(struct f2fs_inode_info
*fi
)
943 if (is_inode_flag_set(fi
, FI_INLINE_XATTR
))
944 return DEF_ADDRS_PER_INODE
- F2FS_INLINE_XATTR_ADDRS
;
945 return DEF_ADDRS_PER_INODE
;
948 static inline void *inline_xattr_addr(struct page
*page
)
950 struct f2fs_inode
*ri
;
951 ri
= (struct f2fs_inode
*)page_address(page
);
952 return (void *)&(ri
->i_addr
[DEF_ADDRS_PER_INODE
-
953 F2FS_INLINE_XATTR_ADDRS
]);
956 static inline int inline_xattr_size(struct inode
*inode
)
958 if (is_inode_flag_set(F2FS_I(inode
), FI_INLINE_XATTR
))
959 return F2FS_INLINE_XATTR_ADDRS
<< 2;
964 static inline int f2fs_readonly(struct super_block
*sb
)
966 return sb
->s_flags
& MS_RDONLY
;
972 int f2fs_sync_file(struct file
*, loff_t
, loff_t
, int);
973 void truncate_data_blocks(struct dnode_of_data
*);
974 void f2fs_truncate(struct inode
*);
975 int f2fs_getattr(struct vfsmount
*, struct dentry
*, struct kstat
*);
976 int f2fs_setattr(struct dentry
*, struct iattr
*);
977 int truncate_hole(struct inode
*, pgoff_t
, pgoff_t
);
978 int truncate_data_blocks_range(struct dnode_of_data
*, int);
979 long f2fs_ioctl(struct file
*, unsigned int, unsigned long);
980 long f2fs_compat_ioctl(struct file
*, unsigned int, unsigned long);
985 void f2fs_set_inode_flags(struct inode
*);
986 struct inode
*f2fs_iget(struct super_block
*, unsigned long);
987 int try_to_free_nats(struct f2fs_sb_info
*, int);
988 void update_inode(struct inode
*, struct page
*);
989 int update_inode_page(struct inode
*);
990 int f2fs_write_inode(struct inode
*, struct writeback_control
*);
991 void f2fs_evict_inode(struct inode
*);
996 struct dentry
*f2fs_get_parent(struct dentry
*child
);
1001 struct f2fs_dir_entry
*f2fs_find_entry(struct inode
*, struct qstr
*,
1003 struct f2fs_dir_entry
*f2fs_parent_dir(struct inode
*, struct page
**);
1004 ino_t
f2fs_inode_by_name(struct inode
*, struct qstr
*);
1005 void f2fs_set_link(struct inode
*, struct f2fs_dir_entry
*,
1006 struct page
*, struct inode
*);
1007 int update_dent_inode(struct inode
*, const struct qstr
*);
1008 int __f2fs_add_link(struct inode
*, const struct qstr
*, struct inode
*);
1009 void f2fs_delete_entry(struct f2fs_dir_entry
*, struct page
*, struct inode
*);
1010 int f2fs_make_empty(struct inode
*, struct inode
*);
1011 bool f2fs_empty_dir(struct inode
*);
1013 static inline int f2fs_add_link(struct dentry
*dentry
, struct inode
*inode
)
1015 return __f2fs_add_link(dentry
->d_parent
->d_inode
, &dentry
->d_name
,
1022 int f2fs_sync_fs(struct super_block
*, int);
1023 extern __printf(3, 4)
1024 void f2fs_msg(struct super_block
*, const char *, const char *, ...);
1029 f2fs_hash_t
f2fs_dentry_hash(const char *, size_t);
1034 struct dnode_of_data
;
1037 int is_checkpointed_node(struct f2fs_sb_info
*, nid_t
);
1038 void get_node_info(struct f2fs_sb_info
*, nid_t
, struct node_info
*);
1039 int get_dnode_of_data(struct dnode_of_data
*, pgoff_t
, int);
1040 int truncate_inode_blocks(struct inode
*, pgoff_t
);
1041 int truncate_xattr_node(struct inode
*, struct page
*);
1042 int wait_on_node_pages_writeback(struct f2fs_sb_info
*, nid_t
);
1043 int remove_inode_page(struct inode
*);
1044 struct page
*new_inode_page(struct inode
*, const struct qstr
*);
1045 struct page
*new_node_page(struct dnode_of_data
*, unsigned int, struct page
*);
1046 void ra_node_page(struct f2fs_sb_info
*, nid_t
);
1047 struct page
*get_node_page(struct f2fs_sb_info
*, pgoff_t
);
1048 struct page
*get_node_page_ra(struct page
*, int);
1049 void sync_inode_page(struct dnode_of_data
*);
1050 int sync_node_pages(struct f2fs_sb_info
*, nid_t
, struct writeback_control
*);
1051 bool alloc_nid(struct f2fs_sb_info
*, nid_t
*);
1052 void alloc_nid_done(struct f2fs_sb_info
*, nid_t
);
1053 void alloc_nid_failed(struct f2fs_sb_info
*, nid_t
);
1054 void recover_node_page(struct f2fs_sb_info
*, struct page
*,
1055 struct f2fs_summary
*, struct node_info
*, block_t
);
1056 int recover_inode_page(struct f2fs_sb_info
*, struct page
*);
1057 int restore_node_summary(struct f2fs_sb_info
*, unsigned int,
1058 struct f2fs_summary_block
*);
1059 void flush_nat_entries(struct f2fs_sb_info
*);
1060 int build_node_manager(struct f2fs_sb_info
*);
1061 void destroy_node_manager(struct f2fs_sb_info
*);
1062 int __init
create_node_manager_caches(void);
1063 void destroy_node_manager_caches(void);
1068 void f2fs_balance_fs(struct f2fs_sb_info
*);
1069 void f2fs_balance_fs_bg(struct f2fs_sb_info
*);
1070 void invalidate_blocks(struct f2fs_sb_info
*, block_t
);
1071 void clear_prefree_segments(struct f2fs_sb_info
*);
1072 int npages_for_summary_flush(struct f2fs_sb_info
*);
1073 void allocate_new_segments(struct f2fs_sb_info
*);
1074 struct page
*get_sum_page(struct f2fs_sb_info
*, unsigned int);
1075 struct bio
*f2fs_bio_alloc(struct block_device
*, int);
1076 void f2fs_submit_bio(struct f2fs_sb_info
*, enum page_type
, bool);
1077 void f2fs_wait_on_page_writeback(struct page
*, enum page_type
, bool);
1078 void write_meta_page(struct f2fs_sb_info
*, struct page
*);
1079 void write_node_page(struct f2fs_sb_info
*, struct page
*, unsigned int,
1080 block_t
, block_t
*);
1081 void write_data_page(struct inode
*, struct page
*, struct dnode_of_data
*,
1082 block_t
, block_t
*);
1083 void rewrite_data_page(struct f2fs_sb_info
*, struct page
*, block_t
);
1084 void recover_data_page(struct f2fs_sb_info
*, struct page
*,
1085 struct f2fs_summary
*, block_t
, block_t
);
1086 void rewrite_node_page(struct f2fs_sb_info
*, struct page
*,
1087 struct f2fs_summary
*, block_t
, block_t
);
1088 void write_data_summaries(struct f2fs_sb_info
*, block_t
);
1089 void write_node_summaries(struct f2fs_sb_info
*, block_t
);
1090 int lookup_journal_in_cursum(struct f2fs_summary_block
*,
1091 int, unsigned int, int);
1092 void flush_sit_entries(struct f2fs_sb_info
*);
1093 int build_segment_manager(struct f2fs_sb_info
*);
1094 void destroy_segment_manager(struct f2fs_sb_info
*);
1095 int __init
create_segment_manager_caches(void);
1096 void destroy_segment_manager_caches(void);
1101 struct page
*grab_meta_page(struct f2fs_sb_info
*, pgoff_t
);
1102 struct page
*get_meta_page(struct f2fs_sb_info
*, pgoff_t
);
1103 long sync_meta_pages(struct f2fs_sb_info
*, enum page_type
, long);
1104 int acquire_orphan_inode(struct f2fs_sb_info
*);
1105 void release_orphan_inode(struct f2fs_sb_info
*);
1106 void add_orphan_inode(struct f2fs_sb_info
*, nid_t
);
1107 void remove_orphan_inode(struct f2fs_sb_info
*, nid_t
);
1108 int recover_orphan_inodes(struct f2fs_sb_info
*);
1109 int get_valid_checkpoint(struct f2fs_sb_info
*);
1110 void set_dirty_dir_page(struct inode
*, struct page
*);
1111 void add_dirty_dir_inode(struct inode
*);
1112 void remove_dirty_dir_inode(struct inode
*);
1113 struct inode
*check_dirty_dir_inode(struct f2fs_sb_info
*, nid_t
);
1114 void sync_dirty_dir_inodes(struct f2fs_sb_info
*);
1115 void write_checkpoint(struct f2fs_sb_info
*, bool);
1116 void init_orphan_info(struct f2fs_sb_info
*);
1117 int __init
create_checkpoint_caches(void);
1118 void destroy_checkpoint_caches(void);
1123 int reserve_new_block(struct dnode_of_data
*);
1124 void update_extent_cache(block_t
, struct dnode_of_data
*);
1125 struct page
*find_data_page(struct inode
*, pgoff_t
, bool);
1126 struct page
*get_lock_data_page(struct inode
*, pgoff_t
);
1127 struct page
*get_new_data_page(struct inode
*, struct page
*, pgoff_t
, bool);
1128 int f2fs_readpage(struct f2fs_sb_info
*, struct page
*, block_t
, int);
1129 int do_write_data_page(struct page
*);
1134 int start_gc_thread(struct f2fs_sb_info
*);
1135 void stop_gc_thread(struct f2fs_sb_info
*);
1136 block_t
start_bidx_of_node(unsigned int, struct f2fs_inode_info
*);
1137 int f2fs_gc(struct f2fs_sb_info
*);
1138 void build_gc_manager(struct f2fs_sb_info
*);
1139 int __init
create_gc_caches(void);
1140 void destroy_gc_caches(void);
1145 int recover_fsync_data(struct f2fs_sb_info
*);
1146 bool space_for_roll_forward(struct f2fs_sb_info
*);
1151 #ifdef CONFIG_F2FS_STAT_FS
1152 struct f2fs_stat_info
{
1153 struct list_head stat_list
;
1154 struct f2fs_sb_info
*sbi
;
1155 struct mutex stat_lock
;
1156 int all_area_segs
, sit_area_segs
, nat_area_segs
, ssa_area_segs
;
1157 int main_area_segs
, main_area_sections
, main_area_zones
;
1158 int hit_ext
, total_ext
;
1159 int ndirty_node
, ndirty_dent
, ndirty_dirs
, ndirty_meta
;
1160 int nats
, sits
, fnids
;
1161 int total_count
, utilization
;
1163 unsigned int valid_count
, valid_node_count
, valid_inode_count
;
1164 unsigned int bimodal
, avg_vblocks
;
1165 int util_free
, util_valid
, util_invalid
;
1166 int rsvd_segs
, overp_segs
;
1167 int dirty_count
, node_pages
, meta_pages
;
1168 int prefree_count
, call_count
;
1169 int tot_segs
, node_segs
, data_segs
, free_segs
, free_secs
;
1170 int tot_blks
, data_blks
, node_blks
;
1171 int curseg
[NR_CURSEG_TYPE
];
1172 int cursec
[NR_CURSEG_TYPE
];
1173 int curzone
[NR_CURSEG_TYPE
];
1175 unsigned int segment_count
[2];
1176 unsigned int block_count
[2];
1177 unsigned base_mem
, cache_mem
;
1180 static inline struct f2fs_stat_info
*F2FS_STAT(struct f2fs_sb_info
*sbi
)
1182 return (struct f2fs_stat_info
*)sbi
->stat_info
;
1185 #define stat_inc_call_count(si) ((si)->call_count++)
1186 #define stat_inc_bggc_count(sbi) ((sbi)->bg_gc++)
1187 #define stat_inc_dirty_dir(sbi) ((sbi)->n_dirty_dirs++)
1188 #define stat_dec_dirty_dir(sbi) ((sbi)->n_dirty_dirs--)
1189 #define stat_inc_total_hit(sb) ((F2FS_SB(sb))->total_hit_ext++)
1190 #define stat_inc_read_hit(sb) ((F2FS_SB(sb))->read_hit_ext++)
1191 #define stat_inc_seg_type(sbi, curseg) \
1192 ((sbi)->segment_count[(curseg)->alloc_type]++)
1193 #define stat_inc_block_count(sbi, curseg) \
1194 ((sbi)->block_count[(curseg)->alloc_type]++)
1196 #define stat_inc_seg_count(sbi, type) \
1198 struct f2fs_stat_info *si = F2FS_STAT(sbi); \
1200 if (type == SUM_TYPE_DATA) \
1206 #define stat_inc_tot_blk_count(si, blks) \
1207 (si->tot_blks += (blks))
1209 #define stat_inc_data_blk_count(sbi, blks) \
1211 struct f2fs_stat_info *si = F2FS_STAT(sbi); \
1212 stat_inc_tot_blk_count(si, blks); \
1213 si->data_blks += (blks); \
1216 #define stat_inc_node_blk_count(sbi, blks) \
1218 struct f2fs_stat_info *si = F2FS_STAT(sbi); \
1219 stat_inc_tot_blk_count(si, blks); \
1220 si->node_blks += (blks); \
1223 int f2fs_build_stats(struct f2fs_sb_info
*);
1224 void f2fs_destroy_stats(struct f2fs_sb_info
*);
1225 void __init
f2fs_create_root_stats(void);
1226 void f2fs_destroy_root_stats(void);
1228 #define stat_inc_call_count(si)
1229 #define stat_inc_bggc_count(si)
1230 #define stat_inc_dirty_dir(sbi)
1231 #define stat_dec_dirty_dir(sbi)
1232 #define stat_inc_total_hit(sb)
1233 #define stat_inc_read_hit(sb)
1234 #define stat_inc_seg_type(sbi, curseg)
1235 #define stat_inc_block_count(sbi, curseg)
1236 #define stat_inc_seg_count(si, type)
1237 #define stat_inc_tot_blk_count(si, blks)
1238 #define stat_inc_data_blk_count(si, blks)
1239 #define stat_inc_node_blk_count(sbi, blks)
1241 static inline int f2fs_build_stats(struct f2fs_sb_info
*sbi
) { return 0; }
1242 static inline void f2fs_destroy_stats(struct f2fs_sb_info
*sbi
) { }
1243 static inline void __init
f2fs_create_root_stats(void) { }
1244 static inline void f2fs_destroy_root_stats(void) { }
1247 extern const struct file_operations f2fs_dir_operations
;
1248 extern const struct file_operations f2fs_file_operations
;
1249 extern const struct inode_operations f2fs_file_inode_operations
;
1250 extern const struct address_space_operations f2fs_dblock_aops
;
1251 extern const struct address_space_operations f2fs_node_aops
;
1252 extern const struct address_space_operations f2fs_meta_aops
;
1253 extern const struct inode_operations f2fs_dir_inode_operations
;
1254 extern const struct inode_operations f2fs_symlink_inode_operations
;
1255 extern const struct inode_operations f2fs_special_inode_operations
;